It's preferable when you want to fit in with existing Java infrastructure. I can prototype something in Jython and then deploy it on Tomcat just like any other Java application.
Y'know, this bug report deals with running Django's test suite, which is really a worst-case scenario. The test suite loads hundreds -- even thousands -- of models into memory, and does many complicated and costly operations. It's a unit test suite, and that makes its performance characteristics relatively unrealistic.
So I wouldn't want to judge real-world performance based on the performance of the test suite.
Also, beta0 is old news at this point; I've got to imagine that the Jython folks squished all sorts of bugs between then and now. Have you given your app another try on the final?
Hey, Jacob Kaplan-Moss, you don't have time to post replies on HN. Django 1.1 is two months late, and I still see 20 open tickets. And would a status update on the Django blog kill you? Also, while you're at it, I've been hearing a lot about the GIL on reddit and HN lately, can you fix that? Thanks.
It's interesting that CPython has picked up Jython/Java conventions for what's "right" over time. That list is from the Python 2.0 era (pre-2002), and most of the items that say CPython should be fixed have been fixed since then -- even down to print becoming a function in Python 3.0 to drop a troublesome keyword.
Python 2.5 is a good baseline, since that's what a lot of distributors are shipping.
Besides, the syntactic additions made to the language in Python 2.6 are almost all meant to provide a first step toward Python 3.0, or make standard things that were available as future imports in 2.5. So you're not really missing anything there.
I remember a post that CPython 2.6 was still in development when they started Jython 2.5. They were making the leap from 2.1 or 2.2 so they didnt want to add an undefined release to the pot.
Can anyone tell me where they find it preferable to use Jython instead of cPython? Also, is there anybody running this for production?